llvm/llvm/test/Transforms/Attributor/IPConstantProp/openmp_parallel_for.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --check-attributes --check-globals
; RUN: opt -aa-pipeline=basic-aa -passes=attributor -attributor-manifest-internal  -attributor-annotate-decl-cs  -S < %s | FileCheck %s --check-prefixes=CHECK,TUNIT
; RUN: opt -aa-pipeline=basic-aa -passes=attributor-cgscc -attributor-manifest-internal  -attributor-annotate-decl-cs -S < %s | FileCheck %s --check-prefixes=CHECK,CGSCC
;
;    void bar(int, float, double);
;
;    void foo(int N) {
;      float p = 3;
;      double q = 5;
;      N = 7;
;
;    #pragma omp parallel for firstprivate(q)
;      for (int i = 2; i < N; i++) {
;        bar(i, p, q);
;      }
;    }
;
; Verify the constant value of q is propagated into the outlined function.
;
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"

%struct.ident_t = type { i32, i32, i32, i32, ptr }

@.str = private unnamed_addr constant [23 x i8] c";unknown;unknown;0;0;;\00", align 1
@0 = private unnamed_addr global %struct.ident_t { i32 0, i32 514, i32 0, i32 0, ptr @.str }, align 8
@1 = private unnamed_addr global %struct.ident_t { i32 0, i32 2, i32 0, i32 0, ptr @.str }, align 8

;.
; CHECK: @.str = private unnamed_addr constant [23 x i8] c"
; CHECK: @[[GLOB0:[0-9]+]] = private unnamed_addr global %struct.ident_t { i32 0, i32 514, i32 0, i32 0, ptr @.str }, align 8
; CHECK: @[[GLOB1:[0-9]+]] = private unnamed_addr global %struct.ident_t { i32 0, i32 2, i32 0, i32 0, ptr @.str }, align 8
;.
define dso_local void @foo(i32 %N) {
; TUNIT-LABEL: define {{[^@]+}}@foo
; TUNIT-SAME: (i32 [[N:%.*]]) {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
; TUNIT-NEXT:    [[P:%.*]] = alloca float, align 4
; TUNIT-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 3, ptr noundef nonnull @.omp_outlined., ptr noalias nocapture nofree nonnull readnone align 4 dereferenceable(4) undef, ptr noalias nocapture nofree nonnull readnone align 4 dereferenceable(4) undef, i64 undef)
; TUNIT-NEXT:    ret void
;
; CGSCC-LABEL: define {{[^@]+}}@foo
; CGSCC-SAME: (i32 [[N:%.*]]) {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    [[N_ADDR:%.*]] = alloca i32, align 4
; CGSCC-NEXT:    [[P:%.*]] = alloca float, align 4
; CGSCC-NEXT:    store i32 [[N]], ptr [[N_ADDR]], align 4
; CGSCC-NEXT:    store float 3.000000e+00, ptr [[P]], align 4
; CGSCC-NEXT:    store i32 7, ptr [[N_ADDR]], align 4
; CGSCC-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr noundef nonnull align 8 dereferenceable(24) @[[GLOB1]], i32 noundef 3, ptr noundef nonnull @.omp_outlined., ptr noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[N_ADDR]], ptr noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P]], i64 noundef 4617315517961601024)
; CGSCC-NEXT:    ret void
;
entry:
  %N.addr = alloca i32, align 4
  %p = alloca float, align 4
  store i32 %N, ptr %N.addr, align 4
  store float 3.000000e+00, ptr %p, align 4
  store i32 7, ptr %N.addr, align 4
  call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr nonnull @1, i32 3, ptr @.omp_outlined., ptr nonnull %N.addr, ptr nonnull %p, i64 4617315517961601024)
  ret void
}

define internal void @.omp_outlined.(ptr noalias %.global_tid., ptr noalias %.bound_tid., ptr dereferenceable(4) %N, ptr dereferenceable(4) %p, i64 %q) {
; TUNIT-LABEL: define {{[^@]+}}@.omp_outlined.
; TUNIT-SAME: (ptr noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], ptr noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], ptr noalias nocapture nofree noundef nonnull readnone align 4 dereferenceable(4) [[N:%.*]], ptr noalias nocapture nofree noundef nonnull readnone align 4 dereferenceable(4) [[P:%.*]], i64 [[Q:%.*]]) {
; TUNIT-NEXT:  entry:
; TUNIT-NEXT:    [[Q_ADDR:%.*]] = alloca i64, align 8
; TUNIT-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
; TUNIT-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
; TUNIT-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
; TUNIT-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
; TUNIT-NEXT:    store i64 4617315517961601024, ptr [[Q_ADDR]], align 8
; TUNIT-NEXT:    br label [[OMP_PRECOND_THEN:%.*]]
; TUNIT:       omp.precond.then:
; TUNIT-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
; TUNIT-NEXT:    store i32 4, ptr [[DOTOMP_UB]], align 4
; TUNIT-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
; TUNIT-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
; TUNIT-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTGLOBAL_TID_]], align 4
; TUNIT-NEXT:    call void @__kmpc_for_static_init_4(ptr noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 [[TMP5]], i32 noundef 34, ptr noundef nonnull align 4 dereferenceable(4) [[DOTOMP_IS_LAST]], ptr noundef nonnull align 4 dereferenceable(4) [[DOTOMP_LB]], ptr noundef nonnull align 4 dereferenceable(4) [[DOTOMP_UB]], ptr noundef nonnull align 4 dereferenceable(4) [[DOTOMP_STRIDE]], i32 noundef 1, i32 noundef 1)
; TUNIT-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
; TUNIT-NEXT:    [[CMP6:%.*]] = icmp sgt i32 [[TMP6]], 4
; TUNIT-NEXT:    br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
; TUNIT:       cond.true:
; TUNIT-NEXT:    br label [[COND_END:%.*]]
; TUNIT:       cond.false:
; TUNIT-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
; TUNIT-NEXT:    br label [[COND_END]]
; TUNIT:       cond.end:
; TUNIT-NEXT:    [[COND:%.*]] = phi i32 [ 4, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
; TUNIT-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
; TUNIT-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
; TUNIT-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
; TUNIT:       omp.inner.for.cond:
; TUNIT-NEXT:    [[DOTOMP_IV_0:%.*]] = phi i32 [ [[TMP8]], [[COND_END]] ], [ [[ADD11:%.*]], [[OMP_INNER_FOR_INC:%.*]] ]
; TUNIT-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
; TUNIT-NEXT:    [[CMP8:%.*]] = icmp sgt i32 [[DOTOMP_IV_0]], [[TMP9]]
; TUNIT-NEXT:    br i1 [[CMP8]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]], label [[OMP_INNER_FOR_BODY:%.*]]
; TUNIT:       omp.inner.for.cond.cleanup:
; TUNIT-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
; TUNIT:       omp.inner.for.body:
; TUNIT-NEXT:    [[ADD10:%.*]] = add nsw i32 [[DOTOMP_IV_0]], 2
; TUNIT-NEXT:    [[TMP11:%.*]] = load double, ptr [[Q_ADDR]], align 8
; TUNIT-NEXT:    call void @bar(i32 [[ADD10]], float nofpclass(nan inf zero sub nnorm) 3.000000e+00, double [[TMP11]])
; TUNIT-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
; TUNIT:       omp.body.continue:
; TUNIT-NEXT:    br label [[OMP_INNER_FOR_INC]]
; TUNIT:       omp.inner.for.inc:
; TUNIT-NEXT:    [[ADD11]] = add nsw i32 [[DOTOMP_IV_0]], 1
; TUNIT-NEXT:    br label [[OMP_INNER_FOR_COND]]
; TUNIT:       omp.inner.for.end:
; TUNIT-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
; TUNIT:       omp.loop.exit:
; TUNIT-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTGLOBAL_TID_]], align 4
; TUNIT-NEXT:    call void @__kmpc_for_static_fini(ptr noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 [[TMP12]])
; TUNIT-NEXT:    br label [[OMP_PRECOND_END:%.*]]
; TUNIT:       omp.precond.end:
; TUNIT-NEXT:    ret void
;
; CGSCC-LABEL: define {{[^@]+}}@.omp_outlined.
; CGSCC-SAME: (ptr noalias nocapture nofree readonly [[DOTGLOBAL_TID_:%.*]], ptr noalias nocapture nofree readnone [[DOTBOUND_TID_:%.*]], ptr noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[N:%.*]], ptr noalias nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[P:%.*]], i64 [[Q:%.*]]) {
; CGSCC-NEXT:  entry:
; CGSCC-NEXT:    [[Q_ADDR:%.*]] = alloca i64, align 8
; CGSCC-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
; CGSCC-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
; CGSCC-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
; CGSCC-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
; CGSCC-NEXT:    store i64 4617315517961601024, ptr [[Q_ADDR]], align 8
; CGSCC-NEXT:    [[TMP:%.*]] = load i32, ptr [[N]], align 4
; CGSCC-NEXT:    [[SUB3:%.*]] = add nsw i32 [[TMP]], -3
; CGSCC-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP]], 2
; CGSCC-NEXT:    br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]
; CGSCC:       omp.precond.then:
; CGSCC-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
; CGSCC-NEXT:    store i32 [[SUB3]], ptr [[DOTOMP_UB]], align 4
; CGSCC-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
; CGSCC-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
; CGSCC-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTGLOBAL_TID_]], align 4
; CGSCC-NEXT:    call void @__kmpc_for_static_init_4(ptr noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 [[TMP5]], i32 noundef 34, ptr noundef nonnull align 4 dereferenceable(4) [[DOTOMP_IS_LAST]], ptr noundef nonnull align 4 dereferenceable(4) [[DOTOMP_LB]], ptr noundef nonnull align 4 dereferenceable(4) [[DOTOMP_UB]], ptr noundef nonnull align 4 dereferenceable(4) [[DOTOMP_STRIDE]], i32 noundef 1, i32 noundef 1)
; CGSCC-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
; CGSCC-NEXT:    [[CMP6:%.*]] = icmp sgt i32 [[TMP6]], [[SUB3]]
; CGSCC-NEXT:    br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
; CGSCC:       cond.true:
; CGSCC-NEXT:    br label [[COND_END:%.*]]
; CGSCC:       cond.false:
; CGSCC-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
; CGSCC-NEXT:    br label [[COND_END]]
; CGSCC:       cond.end:
; CGSCC-NEXT:    [[COND:%.*]] = phi i32 [ [[SUB3]], [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
; CGSCC-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
; CGSCC-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
; CGSCC-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
; CGSCC:       omp.inner.for.cond:
; CGSCC-NEXT:    [[DOTOMP_IV_0:%.*]] = phi i32 [ [[TMP8]], [[COND_END]] ], [ [[ADD11:%.*]], [[OMP_INNER_FOR_INC:%.*]] ]
; CGSCC-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
; CGSCC-NEXT:    [[CMP8:%.*]] = icmp sgt i32 [[DOTOMP_IV_0]], [[TMP9]]
; CGSCC-NEXT:    br i1 [[CMP8]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]], label [[OMP_INNER_FOR_BODY:%.*]]
; CGSCC:       omp.inner.for.cond.cleanup:
; CGSCC-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
; CGSCC:       omp.inner.for.body:
; CGSCC-NEXT:    [[ADD10:%.*]] = add nsw i32 [[DOTOMP_IV_0]], 2
; CGSCC-NEXT:    [[TMP10:%.*]] = load float, ptr [[P]], align 4
; CGSCC-NEXT:    [[TMP11:%.*]] = load double, ptr [[Q_ADDR]], align 8
; CGSCC-NEXT:    call void @bar(i32 [[ADD10]], float [[TMP10]], double [[TMP11]])
; CGSCC-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
; CGSCC:       omp.body.continue:
; CGSCC-NEXT:    br label [[OMP_INNER_FOR_INC]]
; CGSCC:       omp.inner.for.inc:
; CGSCC-NEXT:    [[ADD11]] = add nsw i32 [[DOTOMP_IV_0]], 1
; CGSCC-NEXT:    br label [[OMP_INNER_FOR_COND]]
; CGSCC:       omp.inner.for.end:
; CGSCC-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
; CGSCC:       omp.loop.exit:
; CGSCC-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTGLOBAL_TID_]], align 4
; CGSCC-NEXT:    call void @__kmpc_for_static_fini(ptr noundef nonnull align 8 dereferenceable(24) @[[GLOB0]], i32 [[TMP12]])
; CGSCC-NEXT:    br label [[OMP_PRECOND_END]]
; CGSCC:       omp.precond.end:
; CGSCC-NEXT:    ret void
;
entry:
  %q.addr = alloca i64, align 8
  %.omp.lb = alloca i32, align 4
  %.omp.ub = alloca i32, align 4
  %.omp.stride = alloca i32, align 4
  %.omp.is_last = alloca i32, align 4
  store i64 %q, ptr %q.addr, align 8
  %tmp = load i32, ptr %N, align 4
  %sub3 = add nsw i32 %tmp, -3
  %cmp = icmp sgt i32 %tmp, 2
  br i1 %cmp, label %omp.precond.then, label %omp.precond.end

omp.precond.then:                                 ; preds = %entry
  store i32 0, ptr %.omp.lb, align 4
  store i32 %sub3, ptr %.omp.ub, align 4
  store i32 1, ptr %.omp.stride, align 4
  store i32 0, ptr %.omp.is_last, align 4
  %tmp5 = load i32, ptr %.global_tid., align 4
  call void @__kmpc_for_static_init_4(ptr nonnull @0, i32 %tmp5, i32 34, ptr nonnull %.omp.is_last, ptr nonnull %.omp.lb, ptr nonnull %.omp.ub, ptr nonnull %.omp.stride, i32 1, i32 1)
  %tmp6 = load i32, ptr %.omp.ub, align 4
  %cmp6 = icmp sgt i32 %tmp6, %sub3
  br i1 %cmp6, label %cond.true, label %cond.false

cond.true:                                        ; preds = %omp.precond.then
  br label %cond.end

cond.false:                                       ; preds = %omp.precond.then
  %tmp7 = load i32, ptr %.omp.ub, align 4
  br label %cond.end

cond.end:                                         ; preds = %cond.false, %cond.true
  %cond = phi i32 [ %sub3, %cond.true ], [ %tmp7, %cond.false ]
  store i32 %cond, ptr %.omp.ub, align 4
  %tmp8 = load i32, ptr %.omp.lb, align 4
  br label %omp.inner.for.cond

omp.inner.for.cond:                               ; preds = %omp.inner.for.inc, %cond.end
  %.omp.iv.0 = phi i32 [ %tmp8, %cond.end ], [ %add11, %omp.inner.for.inc ]
  %tmp9 = load i32, ptr %.omp.ub, align 4
  %cmp8 = icmp sgt i32 %.omp.iv.0, %tmp9
  br i1 %cmp8, label %omp.inner.for.cond.cleanup, label %omp.inner.for.body

omp.inner.for.cond.cleanup:                       ; preds = %omp.inner.for.cond
  br label %omp.inner.for.end

omp.inner.for.body:                               ; preds = %omp.inner.for.cond
  %add10 = add nsw i32 %.omp.iv.0, 2
  %tmp10 = load float, ptr %p, align 4
  %tmp11 = load double, ptr %q.addr, align 8
  call void @bar(i32 %add10, float %tmp10, double %tmp11)
  br label %omp.body.continue

omp.body.continue:                                ; preds = %omp.inner.for.body
  br label %omp.inner.for.inc

omp.inner.for.inc:                                ; preds = %omp.body.continue
  %add11 = add nsw i32 %.omp.iv.0, 1
  br label %omp.inner.for.cond

omp.inner.for.end:                                ; preds = %omp.inner.for.cond.cleanup
  br label %omp.loop.exit

omp.loop.exit:                                    ; preds = %omp.inner.for.end
  %tmp12 = load i32, ptr %.global_tid., align 4
  call void @__kmpc_for_static_fini(ptr nonnull @0, i32 %tmp12)
  br label %omp.precond.end

omp.precond.end:                                  ; preds = %omp.loop.exit, %entry
  ret void
}

declare dso_local void @__kmpc_for_static_init_4(ptr, i32, i32, ptr, ptr, ptr, ptr, i32, i32)

declare dso_local void @bar(i32, float, double)

declare dso_local void @__kmpc_for_static_fini(ptr, i32)

declare !callback !0 dso_local void @__kmpc_fork_call(ptr, i32, ptr, ...)

!1 = !{i64 2, i64 -1, i64 -1, i1 true}
!0 = !{!1}
;.
; TUNIT: [[META0:![0-9]+]] = !{[[META1:![0-9]+]]}
; TUNIT: [[META1]] = !{i64 2, i64 -1, i64 -1, i1 true}
;.
; CGSCC: [[META0:![0-9]+]] = !{[[META1:![0-9]+]]}
; CGSCC: [[META1]] = !{i64 2, i64 -1, i64 -1, i1 true}
;.
;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
; CHECK: {{.*}}